problem stringclasses 36
values | Checkpoint number int64 1 8 | language stringclasses 1
value | difficulty stringclasses 3
values | tags listlengths 7 15 | instruction stringlengths 515 15.3k | tests stringlengths 5.59k 5.82k |
|---|---|---|---|---|---|---|
layered_config_synthesizer | 1 | cpp | Medium | [
"cli-tools",
"yaml",
"json",
"mlops",
"merge-algorithms",
"schema-validation",
"template-expansion",
"env-interpolation",
"error-handling",
"file-io"
] | # Part 1 — Core Layering Fundamentals
You are building `synth_configs`, a CLI that turns layered YAML/JSON inputs into a canonical experiment spec for an internal training scheduler. In this part you only build **single-run** configs: merge a base file, required defaults, and an optional override, then surface validat... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for layered_config_synthesizer / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain synth_configs.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkou... |
layered_config_synthesizer | 2 | cpp | Medium | [
"cli-tools",
"yaml",
"json",
"mlops",
"merge-algorithms",
"schema-validation",
"template-expansion",
"env-interpolation",
"error-handling",
"file-io"
] | # Part 2 — Fragments & Environment Interpolation
You are extending `synth_configs` to understand shared fragments and `${VAR}` interpolation. The core merge/validation logic from Part 1 still applies, honor fragment metadata, and substitute environment variables sourced from `env.list`.
## Deliverables
Extend `synth... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for layered_config_synthesizer / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain synth_configs.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkou... |
layered_config_synthesizer | 3 | cpp | Medium | [
"cli-tools",
"yaml",
"json",
"mlops",
"merge-algorithms",
"schema-validation",
"template-expansion",
"env-interpolation",
"error-handling",
"file-io"
] | # Part 3 — Run Manifest & Conflict Surfacing
You are upgrading `synth_configs` to fan out multiple experiment runs from a single manifest. Each run may reference its own overrides and flag files, but all runs share the same base + fragment stack. The tool must emit per-run specs, aggregate summary metrics, and capture... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for layered_config_synthesizer / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain synth_configs.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkou... |
layered_config_synthesizer | 4 | cpp | Medium | [
"cli-tools",
"yaml",
"json",
"mlops",
"merge-algorithms",
"schema-validation",
"template-expansion",
"env-interpolation",
"error-handling",
"file-io"
] | # Part 4 — Schema Validation & Hardware Policies
You are finishing `synth_configs` by enforcing hardware profiles and JSON Schema validation across every run. The tool now ingests declarative schemas plus `hardware.yaml` describing instance capabilities, injects profile defaults, and records schema violations per run.... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for layered_config_synthesizer / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain synth_configs.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkou... |
log_query | 1 | cpp | Medium | [
"query-language",
"parsing",
"ndjson",
"sql",
"aggregation",
"joins",
"cli-tools",
"data-processing",
"schema-mapping",
"subqueries"
] | # LogQL — Part 1 (Parsing, Filtering, Projection)
A command-line program that reads newline-delimited JSON (NDJSON) logs, parses a tiny SQL-like query, filters rows with boolean logic, and returns only the selected fields as JSON.
- **Single table:** Queries operate on exactly one logical table named `logs`, which ma... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for log_query / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain logql.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the ... |
log_query | 2 | cpp | Medium | [
"query-language",
"parsing",
"ndjson",
"sql",
"aggregation",
"joins",
"cli-tools",
"data-processing",
"schema-mapping",
"subqueries"
] | # LogQL — Part 2 (Grouping & Aggregations)
Extend the Part 1 program to support `GROUP BY` and aggregate functions. Results remain **deterministic** and machine-checkable.
- **Single table:** Still only `logs`.
- **No ordering beyond rules below.** No HAVING/ORDER BY/expressions-in-SELECT in this part.
## CLI (uncha... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for log_query / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain logql.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the ... |
log_query | 3 | cpp | Medium | [
"query-language",
"parsing",
"ndjson",
"sql",
"aggregation",
"joins",
"cli-tools",
"data-processing",
"schema-mapping",
"subqueries"
] | # LogQL — Part 3 (Multi-Source Conflation / Joins)
Add **multi-source querying** by binding multiple NDJSON files to **aliases** and querying with a join-like feature using uncommon keywords. You start from one alias in `FROM` and **CONFLATE** others via explicit equality predicates. Aggregation & grouping from Part 2... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for log_query / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain logql.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the ... |
log_query | 4 | cpp | Medium | [
"query-language",
"parsing",
"ndjson",
"sql",
"aggregation",
"joins",
"cli-tools",
"data-processing",
"schema-mapping",
"subqueries"
] | # LogQL — Part 4 (Schema Reconciliation with `GLOSS` / Canonical Labels)
Introduce **canonical labels** so queries stay compact when sources disagree on field names/nesting. A `GLOSS { … }` block maps a **canonical name** to an ordered list of **candidate paths** (optionally with a `DEFAULT`).
## CLI (unchanged)
```... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for log_query / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain logql.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the ... |
log_query | 5 | cpp | Medium | [
"query-language",
"parsing",
"ndjson",
"sql",
"aggregation",
"joins",
"cli-tools",
"data-processing",
"schema-mapping",
"subqueries"
] | # LogQL — Part 5 (Nested Queries & Correlated Subqueries)
Add nested queries as expressions with uncommon keywords:
- **Scalar subquery:** `POCKET(<query>)`
- **Table subquery:** `POCKET[<query>]`
- **Predicates built on tables:** `BEHOLDS` (EXISTS), `AMONGST` (IN), `EITHERWISE` (ANY), `EVERYWISE` (ALL)
- **Correlati... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for log_query / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain logql.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the ... |
meshctl | 1 | cpp | Hard | [
"yaml",
"json",
"resource-management",
"schema-validation",
"state-persistence",
"argument-parsing",
"crud",
"declarative-config",
"distributed-systems"
] | ## meshctl
`meshctl.cpp` manages mesh resources from YAML specs and
prints JSON to stdout.
## Entry Point
```text
./meshctl mesh <operation> [arguments]
```
Entry file: `meshctl.cpp`.
## Commands
| Command | Result |
|---|---|
| `./meshctl mesh create -f <path>` | Read YAML, apply defaults, validate, persist, pri... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for meshctl / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain meshctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the ... |
meshctl | 2 | cpp | Hard | [
"yaml",
"json",
"resource-management",
"schema-validation",
"state-persistence",
"argument-parsing",
"crud",
"declarative-config",
"distributed-systems"
] | ## Mesh Lifecycle and Topology
| Command | Result |
|---|---|
| `./meshctl mesh create -f <path>` | Create a mesh from YAML |
| `./meshctl mesh list` | List mesh summaries |
| `./meshctl mesh describe <name>` | Print the full mesh |
| `./meshctl mesh delete <name>` | Delete the mesh |
| `./meshctl mesh update -f <path... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for meshctl / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain meshctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the ... |
meshctl | 3 | cpp | Hard | [
"yaml",
"json",
"resource-management",
"schema-validation",
"state-persistence",
"argument-parsing",
"crud",
"declarative-config",
"distributed-systems"
] | ## Vault Resource Management
The tool also manages a second resource kind: `vault`.
## Vault Commands
| Command | Result |
|---|---|
| `./meshctl vault create -f <path>` | Read YAML, apply defaults, validate, persist, print the full vault |
| `./meshctl vault list` | Print vault summaries sorted by `name` ascending ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for meshctl / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain meshctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the ... |
meshctl | 4 | cpp | Hard | [
"yaml",
"json",
"resource-management",
"schema-validation",
"state-persistence",
"argument-parsing",
"crud",
"declarative-config",
"distributed-systems"
] | ## meshctl Security Model
`spec.access` supports authentication settings, permission roles, and encryption.
## Authentication
| Field | Type | Default | Rules |
|---|---|---|---|
| `spec.access.authentication.enabled` | boolean | `true` | |
| `spec.access.authentication.digestAlgorithm` | string | `"SHA-256"` | One ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for meshctl / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain meshctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the ... |
meshctl | 5 | cpp | Hard | [
"yaml",
"json",
"resource-management",
"schema-validation",
"state-persistence",
"argument-parsing",
"crud",
"declarative-config",
"distributed-systems"
] | ## One-Shot Operations
The tool also manages three additional resource kinds: `task`, `snapshot`, and `recovery`.
- Creation always starts at `"Initializing"`. Terminal states are irreversible.
- `metadata.name` for all three kinds uses the same validation and not-found shape as mesh resources unless stated otherwise... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for meshctl / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain meshctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the ... |
meshctl | 6 | cpp | Hard | [
"yaml",
"json",
"resource-management",
"schema-validation",
"state-persistence",
"argument-parsing",
"crud",
"declarative-config",
"distributed-systems"
] | ## meshctl Migration Strategies
The tool supports runtime catalog validation, warnings, migration stages, and `mesh migrate`.
## Runtime Catalog
`spec.runtime` remains optional. When present, it must be one of these versions:
| Version | Status |
|---|---|
| `3.0.0` | deprecated |
| `3.1.0` | skipped |
| `3.1.1` | ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for meshctl / checkpoint_6.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain meshctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the ... |
meshctl | 7 | cpp | Hard | [
"yaml",
"json",
"resource-management",
"schema-validation",
"state-persistence",
"argument-parsing",
"crud",
"declarative-config",
"distributed-systems"
] | ## Network Exposure and Connectivity
The tool supports exposure settings, computed connection details, a management endpoint flag, and `mesh shell`.
## Exposure
- `spec.exposure` is optional.
- If `spec.exposure` is omitted:
- no external access is configured
- `status.connectionDetails` is absent
## Exposure T... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for meshctl / checkpoint_7.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain meshctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the ... |
meshctl | 8 | cpp | Hard | [
"yaml",
"json",
"resource-management",
"schema-validation",
"state-persistence",
"argument-parsing",
"crud",
"declarative-config",
"distributed-systems"
] | ## Multi-Region Topology and Operational Policies
The tool supports region topology, telemetry tags, placement defaults, config bundle refresh tracking, and extensions.
## Always-Present Output Additions
Create and describe output for every mesh now includes:
| Path | Rules |
|---|---|
| `spec.placement` | Present ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for meshctl / checkpoint_8.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain meshctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the ... |
metric_transform_lang | 1 | cpp | Hard | [
"dsl",
"interpreter",
"data-processing",
"streaming",
"json",
"ndjson",
"aggregation",
"window-functions",
"schema-validation",
"multi-format-io",
"parquet",
"deterministic-execution",
"stateful-processing",
"relational-joins",
"type-system"
] | A retailer streams raw sales events and wants reliable ad-hoc rollups without shipping code. You'll implement a **file-based CLI** that reads a CSV of events and a small **domain specific language (DSL)** describing a transformation pipeline, then emits **deterministic JSON** according to the rules below.
The DSL’s op... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for metric_transform_lang / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain mtl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cl... |
metric_transform_lang | 2 | cpp | Hard | [
"dsl",
"interpreter",
"data-processing",
"streaming",
"json",
"ndjson",
"aggregation",
"window-functions",
"schema-validation",
"multi-format-io",
"parquet",
"deterministic-execution",
"stateful-processing",
"relational-joins",
"type-system"
] | The data platform team wants reusable, evolvable metric pipelines shared across teams. You’ll extend your Part 1 CLI to parse a richer DSL with **schema blocks** that support **inheritance and type overrides**, **calculated fields** (pre- and post-aggregation), **pipeline parameters**, and **post-aggregation LAG/LEAD**... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for metric_transform_lang / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain mtl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cl... |
metric_transform_lang | 3 | cpp | Hard | [
"dsl",
"interpreter",
"data-processing",
"streaming",
"json",
"ndjson",
"aggregation",
"window-functions",
"schema-validation",
"multi-format-io",
"parquet",
"deterministic-execution",
"stateful-processing",
"relational-joins",
"type-system"
] | You'll extend the same CLI to:
1. ingest **multiple files** (single file or directory, optional recursion),
2. handle **different formats** (CSV, CSV.GZ, TSV, TSV.GZ, Parquet),
3. be **order-independent** (identical results regardless of input sequence), and
4. deal with **missing values and corrupted data** determini... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for metric_transform_lang / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain mtl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cl... |
metric_transform_lang | 4 | cpp | Hard | [
"dsl",
"interpreter",
"data-processing",
"streaming",
"json",
"ndjson",
"aggregation",
"window-functions",
"schema-validation",
"multi-format-io",
"parquet",
"deterministic-execution",
"stateful-processing",
"relational-joins",
"type-system"
] | The data team wants one pipeline to stitch multiple upstream feeds (web, POS, promos), validate/convert per-file schemas, and compute cross-stream metrics with custom logic. You’ll extend your CLI to load ingest **multiple named sources** configured by a **declarative ingest config**, **join** sources, then run the nor... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for metric_transform_lang / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain mtl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cl... |
metric_transform_lang | 5 | cpp | Hard | [
"dsl",
"interpreter",
"data-processing",
"streaming",
"json",
"ndjson",
"aggregation",
"window-functions",
"schema-validation",
"multi-format-io",
"parquet",
"deterministic-execution",
"stateful-processing",
"relational-joins",
"type-system"
] | Analysts want stable, evolvable pipelines they can pause/retry and audit. Extend your CLI to:
1. **Gate features by DSL version** (explicit version header + compatibility rules),
2. **Resume** deterministically from a prior run (idempotent, exactly-once per file unit),
3. **Emit structured metadata** alongside results... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for metric_transform_lang / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain mtl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cl... |
migrate_configs | 1 | cpp | Easy | [
"file-io",
"config-management",
"rules-engine",
"pattern-matching",
"data-transformation",
"multi-format-parsing",
"json",
"yaml"
] | You're building a CLI tool to help migrate configuration files across different environments. Your tool will apply transformation rules to a directory tree of config files in various formats.
## Task
Implement a CLI tool that applies transformation rules from a `rules.json` file to all config files in a target direct... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for migrate_configs / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain migrate.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# clon... |
migrate_configs | 2 | cpp | Easy | [
"file-io",
"config-management",
"rules-engine",
"pattern-matching",
"data-transformation",
"multi-format-parsing",
"json",
"yaml"
] | # Config Migration Tool - Checkpoint 2
Extend your migration tool to support pattern matching, string templates, and conditional replacements.
## New Rule Types
### `pattern_replace`
Match keys using patterns with variables or wildcards, then replace their values.
```json
{
"type": "pattern_replace",
"key_patt... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for migrate_configs / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain migrate.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# clon... |
migrate_configs | 3 | cpp | Easy | [
"file-io",
"config-management",
"rules-engine",
"pattern-matching",
"data-transformation",
"multi-format-parsing",
"json",
"yaml"
] | # Config Migration Tool - Checkpoint 3
Extend your migration tool to handle nested data structures and config inheritance.
## Array Pattern Matching
Patterns can now traverse arrays using bracket notation.
### Syntax
- `array[*]` - Matches all elements
- `array[0]`, `array[1]` - Matches specific index
- `array[?ke... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for migrate_configs / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain migrate.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# clon... |
migrate_configs | 4 | cpp | Easy | [
"file-io",
"config-management",
"rules-engine",
"pattern-matching",
"data-transformation",
"multi-format-parsing",
"json",
"yaml"
] | # Config Migration Tool - Checkpoint 4
Extend your migration tool to move and rename config files based on path patterns and file contents.
## File Relocation Rules
Your tool can now rename and move config files in addition to transforming their contents. File relocation rules have a special `scope` field set to `"f... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for migrate_configs / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain migrate.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# clon... |
migrate_configs | 5 | cpp | Easy | [
"file-io",
"config-management",
"rules-engine",
"pattern-matching",
"data-transformation",
"multi-format-parsing",
"json",
"yaml"
] | # Config Migration Tool - Checkpoint 5
Add validation rules to ensure config files meet requirements before applying transformations.
## Validation Rules
Validation rules check that configs satisfy constraints. They have a `scope` field set to `"validation"` to distinguish them from transformation and relocation rul... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for migrate_configs / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain migrate.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# clon... |
mocked_http | 1 | cpp | Hard | [
"mock-server",
"yaml",
"http",
"templating",
"redis",
"grpc",
"kafka",
"amqp",
"rest-api"
] | Build a multi-protocol mock server driven by YAML mock definitions. This revision covers HTTP request mocking only.
# Deliverable
Create `hmock.cpp`.
- Start the server with:
```
./hmock
```
# Environment
| Variable | Default | Meaning |
|----------|---------|---------|
| `HM_TEMPLATES_DIR` | `./templates` | Dire... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mocked_http / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain hmock.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
mocked_http | 2 | cpp | Hard | [
"mock-server",
"yaml",
"http",
"templating",
"redis",
"grpc",
"kafka",
"amqp",
"rest-api"
] | Add richer template helpers and file-backed response bodies.
# Template Helpers
The following functions are available in all template contexts.
## `jsonPath(expr, data) -> string`
- Query JSON with XPath-style syntax such as `"foo"` or `"//bar"`.
- Return the matched node's inner text.
- Return `""` when `data` is ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mocked_http / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain hmock.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
mocked_http | 3 | cpp | Hard | [
"mock-server",
"yaml",
"http",
"templating",
"redis",
"grpc",
"kafka",
"amqp",
"rest-api"
] | Add stateful actions through Redis and outbound HTTP side effects.
# Redis Backend
## Environment
| Variable | Default | Meaning |
|----------|---------|---------|
| `HM_REDIS_TYPE` | `memory` | Redis backend type: `memory` or `redis` |
| `HM_REDIS_URL` | `redis://redis:6379` | External Redis URL when `HM_REDIS_TYPE... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mocked_http / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain hmock.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
mocked_http | 4 | cpp | Hard | [
"mock-server",
"yaml",
"http",
"templating",
"redis",
"grpc",
"kafka",
"amqp",
"rest-api"
] | Add reusable templates, inheritance, values, and explicit action ordering.
# Kinds
`kind` now accepts three values:
| Kind | Meaning |
|------|---------|
| `Behavior` | Concrete mock. Matches requests and executes actions |
| `Template` | Reusable template fragment |
| `AbstractBehavior` | Base behavior for inherita... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mocked_http / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain hmock.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
mocked_http | 5 | cpp | Hard | [
"mock-server",
"yaml",
"http",
"templating",
"redis",
"grpc",
"kafka",
"amqp",
"rest-api"
] | Add an admin API, template sets, and persistent storage for API-added mocks.
# Admin Server
## Environment
| Variable | Default | Meaning |
|----------|---------|---------|
| `HM_ADMIN_HTTP_ENABLED` | `true` | Enable or disable the admin server |
| `HM_ADMIN_HTTP_PORT` | `9998` | Admin listen port |
| `HM_ADMIN_HTTP... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mocked_http / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain hmock.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
mocked_http | 6 | cpp | Hard | [
"mock-server",
"yaml",
"http",
"templating",
"redis",
"grpc",
"kafka",
"amqp",
"rest-api"
] | Add hot reload, CORS, binary file payloads, and an admin CLI.
# Hot Reload
## Environment
| Variable | Default | Meaning |
|----------|---------|---------|
| `HM_TEMPLATES_DIR_HOT_RELOAD` | `true` | Make filesystem edits visible automatically |
- When enabled, creations, edits, and deletions under the templates dir... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mocked_http / checkpoint_6.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain hmock.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
mocked_http | 7 | cpp | Hard | [
"mock-server",
"yaml",
"http",
"templating",
"redis",
"grpc",
"kafka",
"amqp",
"rest-api"
] | Add Kafka and AMQP message handling.
# Kafka
## Environment
| Variable | Default | Meaning |
|----------|---------|---------|
| `HM_KAFKA_ENABLED` | `false` | Enable Kafka consume and publish |
| `HM_KAFKA_CLIENT_ID` | `hmock` | Kafka client ID |
| `HM_KAFKA_SEED_BROKERS` | `kafka:9092` | Comma-separated default bro... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mocked_http / checkpoint_7.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain hmock.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
mocked_http | 8 | cpp | Hard | [
"mock-server",
"yaml",
"http",
"templating",
"redis",
"grpc",
"kafka",
"amqp",
"rest-api"
] | Add gRPC mocking and a dry-run evaluation endpoint.
# gRPC
## Environment
| Variable | Default | Meaning |
|----------|---------|---------|
| `HM_GRPC_ENABLED` | `false` | Enable the gRPC server |
| `HM_GRPC_PORT` | `50051` | gRPC listen port |
| `HM_GRPC_HOST` | `0.0.0.0` | gRPC listen address |
| `HM_GRPC_DESCRIPT... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mocked_http / checkpoint_8.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain hmock.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
mvvault | 1 | cpp | Medium | [
"media-archival",
"metadata-tracking",
"http-server",
"json",
"file-io",
"migration",
"crud",
"history-tracking"
] | # mvault
`mvault` creates local vaults for media-platform metadata and records
tracked-field history by sync timestamp.
## `mvault` Entry Points
| Surface | Syntax | Required Result |
|---|---|---|
| Console command | `./mvault <subcommand> [args...]` | Full CLI behavior |
| Entry file | `mvault.cpp` | Required impl... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mvvault / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain mvault.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the p... |
mvvault | 2 | cpp | Medium | [
"media-archival",
"metadata-tracking",
"http-server",
"json",
"file-io",
"migration",
"crud",
"history-tracking"
] | # mvault
`mvault` supports transparent loading of legacy vault catalogs. Older catalog
versions use different schema layouts and timestamp formats. Commands that
read vault data must handle all supported versions without requiring prior
migration.
## Supported Versions
| Version | Status | Shape Summary |
|---|---|-... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mvvault / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain mvault.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the p... |
mvvault | 3 | cpp | Medium | [
"media-archival",
"metadata-tracking",
"http-server",
"json",
"file-io",
"migration",
"crud",
"history-tracking"
] | # mvault
`mvault` also supports selective download behavior during `sync`, a `digest`
command, and a post-sync change summary. The `digest` command works on vaults
of any supported version without requiring prior migration.
## `sync` Command
| Item | Value |
|---|---|
| Syntax | `./mvault sync <name> [options]` |
| ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mvvault / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain mvault.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the p... |
mvvault | 4 | cpp | Medium | [
"media-archival",
"metadata-tracking",
"http-server",
"json",
"file-io",
"migration",
"crud",
"history-tracking"
] | # mvault
`mvault` also provides a local browser viewer and appends viewer links to
human-facing change reports. The viewer works on vaults of any supported
version without requiring prior migration.
## `serve` Command
| Item | Value |
|---|---|
| Syntax | `./mvault serve [<name>] [--host=<host>] [--port=<port>]` |
|... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mvvault / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain mvault.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the p... |
mvvault | 5 | cpp | Medium | [
"media-archival",
"metadata-tracking",
"http-server",
"json",
"file-io",
"migration",
"crud",
"history-tracking"
] | # mvault
The viewer also supports entry detail pages with metadata, charts, and direct
serving of archived media and preview files. Detail pages and charts work on
vaults of any supported version, normalizing across schema differences.
## `/catalog/<name>/<category>/<id>` Route
| Method | Path | Required Behavior |
... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mvvault / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain mvault.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the p... |
mvvault | 6 | cpp | Medium | [
"media-archival",
"metadata-tracking",
"http-server",
"json",
"file-io",
"migration",
"crud",
"history-tracking"
] | # mvault
Entry detail pages support user-managed annotations. Annotations require
version 3 catalog format. Writing an annotation to a v1 or v2 vault triggers
automatic migration to v3 before the annotation is persisted.
## `/catalog/<name>/<category>/<id>` Annotation Methods
| Method | Purpose | Request Body | Succ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for mvvault / checkpoint_6.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain mvault.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the p... |
pwd_manager | 1 | cpp | Medium | [
"interactive",
"encryption",
"secret-management",
"file-io",
"clipboard",
"tab-completion",
"import-export",
"state-management"
] | Build an interactive command-line password manager that stores secrets in an encrypted vault.
You launch a single interactive program with no subcommands:
```
./pwd_manager
```
The program reads all input from standard input. Password prompts are hidden when the session supports that behavior, but piped input still ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for pwd_manager / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain pwd_manager.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# clon... |
pwd_manager | 2 | cpp | Medium | [
"interactive",
"encryption",
"secret-management",
"file-io",
"clipboard",
"tab-completion",
"import-export",
"state-management"
] | The vault needs organization and basic record maintenance. Add categories for grouping secrets, and let the user edit or delete a secret after opening its item view.
The main menu includes search, show all, add, quit, and category commands. Entering `cat` opens category management. That screen clears the terminal, sho... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for pwd_manager / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain pwd_manager.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# clon... |
pwd_manager | 3 | cpp | Medium | [
"interactive",
"encryption",
"secret-management",
"file-io",
"clipboard",
"tab-completion",
"import-export",
"state-management"
] | The vault should be fast to use at the keyboard. Add clipboard actions in item view, a temporary on-screen password reveal, password suggestions during password entry, and tab-completion for search and login prompts.
The item view supports edit, delete, search, and back, and advertises copy and display actions based o... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for pwd_manager / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain pwd_manager.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# clon... |
pwd_manager | 4 | cpp | Medium | [
"interactive",
"encryption",
"secret-management",
"file-io",
"clipboard",
"tab-completion",
"import-export",
"state-management"
] | The vault needs non-interactive file workflows and runtime configuration. Add command-line flags for timeout and path settings, plus one-shot JSON import and export modes.
The program accepts optional flags before entering the interactive flow. Support `-t N` for clipboard auto-clear seconds with a default of 15, `-p ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for pwd_manager / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain pwd_manager.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# clon... |
pwd_manager | 5 | cpp | Medium | [
"interactive",
"encryption",
"secret-management",
"file-io",
"clipboard",
"tab-completion",
"import-export",
"state-management"
] | The vault supports idle-session hardening and recovery workflows. Add automatic locking, a manual lock command, an erase mode, and an advisory for changing the master key.
The vault now locks itself after a period of inactivity, using 900 seconds by default or the current configured value from `-a`. Once the timeout i... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for pwd_manager / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain pwd_manager.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# clon... |
recli | 1 | cpp | Hard | [
"cli-tools",
"configuration",
"yaml",
"argument-parsing",
"sqlite",
"caching",
"docker",
"version-management",
"output-formatting"
] | # CLI Entry Point, Command Tree, and Argument Parsing
## Overview
Build `appctl.cpp`, a command-line tool that dispatches hierarchical commands. The tool must support multi-level command paths such as `site create`, validate the selected command's arguments before running any handler, and report mistakes with clear e... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for recli / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain appctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the pro... |
recli | 2 | cpp | Hard | [
"cli-tools",
"configuration",
"yaml",
"argument-parsing",
"sqlite",
"caching",
"docker",
"version-management",
"output-formatting"
] | # Configuration System, Aliases, and Global Parameters
## Scope
The tool supports configuration loading, alias resolution, and global parameter handling in addition to the command tree.
## Configuration Loading
Configuration values come from three sources. The implementation must resolve them using the following pr... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for recli / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain appctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the pro... |
recli | 3 | cpp | Hard | [
"cli-tools",
"configuration",
"yaml",
"argument-parsing",
"sqlite",
"caching",
"docker",
"version-management",
"output-formatting"
] | # Output Formatting, Logging, and User Interaction
## Multi-Format Output
Command results are rendered according to `--format`. The default is `table`.
| Format | Description |
|--------|-------------|
| `table` | Fixed-width table with widths computed from the widest value in each column |
| `csv` | Comma-delimited... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for recli / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain appctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the pro... |
recli | 4 | cpp | Hard | [
"cli-tools",
"configuration",
"yaml",
"argument-parsing",
"sqlite",
"caching",
"docker",
"version-management",
"output-formatting"
] | # Process Execution, File Cache, and Archive Extraction
## Scope
The tool supports external command execution, cache management, and archive extraction.
## External Command Execution
### Detailed Execution
Command execution must be able to expose command output, exit status, and elapsed runtime.
Before execution,... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for recli / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain appctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the pro... |
recli | 5 | cpp | Hard | [
"cli-tools",
"configuration",
"yaml",
"argument-parsing",
"sqlite",
"caching",
"docker",
"version-management",
"output-formatting"
] | # Persistent State Database
## Scope
The tool maintains a local persistent store used to keep key-value settings across invocations and to track applied schema migrations.
## Database Initialization
The implementation must maintain a persistent store under `<APPCTL_ROOT_DIR>/db/`.
1. On first access, if the store ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for recli / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain appctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the pro... |
recli | 6 | cpp | Hard | [
"cli-tools",
"configuration",
"yaml",
"argument-parsing",
"sqlite",
"caching",
"docker",
"version-management",
"output-formatting"
] | # Container Operations, Volumes, Networks, and Compose Integration
## Compose File Merging
Whenever the tool runs a compose-driven operation, it must combine the base project compose definition with any optional overrides using deterministic precedence.
1. If `custom-compose` is set to a non-empty path, that overrid... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for recli / checkpoint_6.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain appctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the pro... |
recli | 7 | cpp | Hard | [
"cli-tools",
"configuration",
"yaml",
"argument-parsing",
"sqlite",
"caching",
"docker",
"version-management",
"output-formatting"
] | # Version Upgrade Infrastructure
## Scope
The tool supports upgrade orchestration: versioned migrations, sequential image pulls, global and per-project container upgrades, compose tool replacement, rollback, and post-upgrade cleanup.
## Automatic Multi-Step Rollback
When the tool performs a multi-step operation and... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for recli / checkpoint_7.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain appctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the pro... |
recli | 8 | cpp | Hard | [
"cli-tools",
"configuration",
"yaml",
"argument-parsing",
"sqlite",
"caching",
"docker",
"version-management",
"output-formatting"
] | # System Requirements, First-Run Setup, CLI Management, and Tab Completion
## Scope
The tool supports environment validation, first-run setup, startup version checks, auto-update behavior, runtime binary resolution, CLI self-management commands, and tab completion.
## System Requirement Checks
Before most commands,... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for recli / checkpoint_8.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain appctl.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the pro... |
rejector | 1 | cpp | Hard | [
"cli",
"api-client",
"rate-limiting",
"concurrency",
"data-generation",
"llm",
"templates",
"agentic",
"pipeline"
] | Build a CLI tool that reads a YAML task configuration and a JSONL input
file, sends prompts to an OpenAI-compatible chat completions API, and
writes one JSONL result per input row. Part 1 covers a single task with
three generation schemes: `greedy`, `sample`, and `rejection`.
The API server queues requests internally ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for rejector / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain rejector.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
rejector | 2 | cpp | Hard | [
"cli",
"api-client",
"rate-limiting",
"concurrency",
"data-generation",
"llm",
"templates",
"agentic",
"pipeline"
] | Extend Part 1 to support multiple named tasks in one config file. Each task
has its own prompt, generation settings, evaluation, and output field. This
part also adds `llm_judge` and `script` evaluation modes.
Existing behavior from Part 1 is unchanged unless stated here.
# Configuration
Multi-task configs use `defa... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for rejector / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain rejector.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
rejector | 3 | cpp | Hard | [
"cli",
"api-client",
"rate-limiting",
"concurrency",
"data-generation",
"llm",
"templates",
"agentic",
"pipeline"
] | Extend Part 2 with in-context learning (ICL) setups and support for
producing multiple solutions per input. Rejection sampling now aims to
collect multiple passing solutions instead of stopping at the first pass.
Existing behavior from earlier parts is unchanged unless stated here.
# ICL Configuration
Tasks may defi... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for rejector / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain rejector.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
rejector | 4 | cpp | Hard | [
"cli",
"api-client",
"rate-limiting",
"concurrency",
"data-generation",
"llm",
"templates",
"agentic",
"pipeline"
] | Extend the pipeline with an `agentic` generation mode and support for
template-driven requests against `/v1/completions`. Agentic tasks let the
model call tools in a loop before producing a final answer.
Existing behavior from earlier parts is unchanged unless stated here.
# Agentic Generation
Task example:
```yaml... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for rejector / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain rejector.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
rejector | 5 | cpp | Hard | [
"cli",
"api-client",
"rate-limiting",
"concurrency",
"data-generation",
"llm",
"templates",
"agentic",
"pipeline"
] | Extend the pipeline with token-aware scheduling, cost accounting, structured
output validation, resume support, dry-run estimation, and progress
reporting.
Existing behavior from earlier parts is unchanged unless stated here.
# Rate Limits
Configuration:
```yaml
defaults:
rate_limits:
rpm: 100
tpm: 50000
... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for rejector / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain rejector.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
sheeteval | 1 | cpp | Hard | [
"cli-tools",
"spreadsheet",
"xlsx",
"grading",
"formula-evaluation",
"yaml",
"html",
"report-generation",
"deterministic-testing"
] | ## sheeteval
`sheeteval` reads an instructor's answer-key spreadsheet and a student's submission spreadsheet, evaluates the submission against grading rules embedded in the answer key, and reports scores.
## Invocation
```
./sheeteval <answer-key> <submission> [options]
```
**Positional arguments:**
| Argument | D... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for sheeteval / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain sheeteval.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning ... |
sheeteval | 2 | cpp | Hard | [
"cli-tools",
"spreadsheet",
"xlsx",
"grading",
"formula-evaluation",
"yaml",
"html",
"report-generation",
"deterministic-testing"
] | ## Scoring Extensions and Text Reporting
`sheeteval` extends its evaluation model with numeric tolerance, alternate valid answers, and penalty scoring, and gains text report output and detailed streaming output to stdout.
---
### Numeric Tolerance
The `rule.tolerance` field accepts a non-negative float defining an ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for sheeteval / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain sheeteval.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning ... |
sheeteval | 3 | cpp | Hard | [
"cli-tools",
"spreadsheet",
"xlsx",
"grading",
"formula-evaluation",
"yaml",
"html",
"report-generation",
"deterministic-testing"
] | ## sheeteval — Formula-Based Rule Kinds
`sheeteval` gains two formula-aware rule kinds — `expression` and `computed` — along with a formula evaluation engine that resolves cell references, arithmetic, and a fixed set of spreadsheet functions.
### `expression` Kind
The `expression` kind compares the submission cell's... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for sheeteval / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain sheeteval.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning ... |
sheeteval | 4 | cpp | Hard | [
"cli-tools",
"spreadsheet",
"xlsx",
"grading",
"formula-evaluation",
"yaml",
"html",
"report-generation",
"deterministic-testing"
] | ## sheeteval — Prerequisite Gating, Grading Flags, and Score Thresholds
Three control-flow mechanisms modify grading around individual rule evaluation: prerequisite gating, fatal and concealed grading flags, and per-sheet minimum-score thresholds. These mechanisms apply uniformly to all rule kinds and interact with al... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for sheeteval / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain sheeteval.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning ... |
sheeteval | 5 | cpp | Hard | [
"cli-tools",
"spreadsheet",
"xlsx",
"grading",
"formula-evaluation",
"yaml",
"html",
"report-generation",
"deterministic-testing"
] | ## Cross-Reference and Boolean Rule Kinds
`sheeteval` supports four additional rule kinds — `reference`, `reference_expr`, `result`, and `boolean` — that evaluate submission data against the answer key using cross-cell formula verification, target-cell comparison, and boolean assertions.
## `reference` Kind
The `ref... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for sheeteval / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain sheeteval.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning ... |
sheeteval | 6 | cpp | Hard | [
"cli-tools",
"spreadsheet",
"xlsx",
"grading",
"formula-evaluation",
"yaml",
"html",
"report-generation",
"deterministic-testing"
] | Check mode reinterprets rule outcomes as checks while keeping the underlying evaluation logic unchanged.
---
### Inputs and Outputs
**New CLI flag:** `--check-mode` / `-C`.
**New _Checklist columns:** `check-name`, `expected-outcome`, `fail-message`, `expected-points`. All optional.
**New annotation fields:** `sce... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for sheeteval / checkpoint_6.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain sheeteval.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning ... |
sheeteval | 7 | cpp | Hard | [
"cli-tools",
"spreadsheet",
"xlsx",
"grading",
"formula-evaluation",
"yaml",
"html",
"report-generation",
"deterministic-testing"
] | ## HTML Report Output
`sheeteval` supports generating a standalone HTML report via the `--html-file <path>` flag. The HTML report renders grading results as structured tables — one per graded sheet — with subtotals and an assignment total. The same grading pass that produces text reports, score files, and stdout outpu... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for sheeteval / checkpoint_7.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain sheeteval.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning ... |
sith | 1 | cpp | Hard | [
"cli-tools",
"python",
"repl",
"static-analysis",
"interpreter-mode",
"code-intelligence",
"autocompletion",
"type-inference",
"refactoring",
"environment-management",
"project-configuration",
"ast-parsing"
] | Build a command-line Python code intelligence tool that analyzes Python source files and provides autocompletion suggestions at a given cursor position. The tool parses Python code statically, resolves visible names from scopes, and returns ranked completions.
# Deliverables
Write `sith.cpp` with this interface:
```... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for sith / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain sith.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the proble... |
sith | 2 | cpp | Hard | [
"cli-tools",
"python",
"repl",
"static-analysis",
"interpreter-mode",
"code-intelligence",
"autocompletion",
"type-inference",
"refactoring",
"environment-management",
"project-configuration",
"ast-parsing"
] | Add type inference and definition navigation.
---
## New Subcommands
### `infer`
```
./sith infer <file> <line> <col>
```
Returns what the name at the cursor position evaluates to. Output is a JSON object with a `"definitions"` array.
If the cursor is not on a name, exit 1. If the name cannot be resolved, return ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for sith / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain sith.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the proble... |
sith | 3 | cpp | Hard | [
"cli-tools",
"python",
"repl",
"static-analysis",
"interpreter-mode",
"code-intelligence",
"autocompletion",
"type-inference",
"refactoring",
"environment-management",
"project-configuration",
"ast-parsing"
] | Extend the tool to resolve imports across files, navigate to definitions in other modules, and provide import-context completions. The tool now operates on a project rather than a single file in isolation.
---
## Project Root
All commands now accept an optional `--project <dir>` flag. When provided, the project root... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for sith / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain sith.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the proble... |
sith | 4 | cpp | Hard | [
"cli-tools",
"python",
"repl",
"static-analysis",
"interpreter-mode",
"code-intelligence",
"autocompletion",
"type-inference",
"refactoring",
"environment-management",
"project-configuration",
"ast-parsing"
] | Add function signature help, find-references, project-wide search, and stub file support for type information from `.pyi` files.
---
## New Subcommands
### `signatures`
```
./sith signatures <file> <line> <col> [--project <dir>]
```
When the cursor is inside a function call's argument list, return the function's s... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for sith / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain sith.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the proble... |
sith | 5 | cpp | Hard | [
"cli-tools",
"python",
"repl",
"static-analysis",
"interpreter-mode",
"code-intelligence",
"autocompletion",
"type-inference",
"refactoring",
"environment-management",
"project-configuration",
"ast-parsing"
] | Add refactoring operations that transform code safely across files, and a syntax error reporting command.
---
## New Subcommands
### `rename`
```
./sith rename <file> <line> <col> --new-name <name> [--diff] [--project <dir>]
```
Rename the name at the cursor and all its references across the project.
| Field | Ty... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for sith / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain sith.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the proble... |
sith | 6 | cpp | Hard | [
"cli-tools",
"python",
"repl",
"static-analysis",
"interpreter-mode",
"code-intelligence",
"autocompletion",
"type-inference",
"refactoring",
"environment-management",
"project-configuration",
"ast-parsing"
] | Add an interpreter mode for REPL-style analysis with live namespaces, Python environment detection, and persistent project configuration.
---
## Interpreter Mode
All analysis commands (`complete`, `infer`, `goto`, `signatures`) now accept an `--interpreter` flag with a `--namespaces` argument:
### Namespaces File
... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for sith / checkpoint_6.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain sith.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the proble... |
test_translator | 1 | cpp | Hard | [
"code-generation",
"test-harness",
"multi-language",
"python",
"javascript",
"typescript",
"test-frameworks",
"parsing",
"file-io"
] | Write `babel_code_goat.cpp` that implements a test-harness translation engine and runner. Support the following commands:
* `generate <tests_dir> --entrypoint <entrypoint function/class method name> --lang <target_lang> [flags...]`
* `test <solution_path> <tests_dir> --lang <target_lang> [flags...]`
**Notes:**
* `<t... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for test_translator / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain babel_code_goat.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoi... |
test_translator | 2 | cpp | Hard | [
"code-generation",
"test-harness",
"multi-language",
"python",
"javascript",
"typescript",
"test-frameworks",
"parsing",
"file-io"
] | The translation harness supports:
### Dict keys
Python dict keys may be **any valid key type** expressible in tests (not restricted to strings).
### Additional Python “primitive containers”
Python tests may use:
* `set` / `frozenset`
* `collections`:
* `Counter`
* `deque`
* `defaultdict`
* `decimal.Dec... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for test_translator / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain babel_code_goat.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoi... |
test_translator | 3 | cpp | Hard | [
"code-generation",
"test-harness",
"multi-language",
"python",
"javascript",
"typescript",
"test-frameworks",
"parsing",
"file-io"
] | The test harness supports:
### Single-call constraint
Each test must be traceable to exactly one entrypoint invocation. Multiple entrypoint calls in one test are unsupported.
###“No multi-function calls”
Aside from primitive operations and tolerance helpers, tests must not depend on other calls.
### Primitive opera... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for test_translator / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain babel_code_goat.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoi... |
test_translator | 4 | cpp | Hard | [
"code-generation",
"test-harness",
"multi-language",
"python",
"javascript",
"typescript",
"test-frameworks",
"parsing",
"file-io"
] |
For-loops are valid test constructs. The loop statement itself is a test, and assertions inside the loop body follow single-call traceability rules.
---
## Loop-as-Test
A parameterization loop statement itself is a test:
* **pass** if it iterates at least once
* **fail** if zero iterations (or cannot be evaluated)... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for test_translator / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain babel_code_goat.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoi... |
test_translator | 5 | cpp | Hard | [
"code-generation",
"test-harness",
"multi-language",
"python",
"javascript",
"typescript",
"test-frameworks",
"parsing",
"file-io"
] | Mutation-style tests with explicit constraints:
### Mutation-style requirements
Mutation-style tests are only valid when:
* an entrypoint call is a statement or assignment, and
* immediately followed by one or more asserts with no entrypoint call, and
* each such assert references at least one variable passed to the... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for test_translator / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain babel_code_goat.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoi... |
test_translator | 6 | cpp | Hard | [
"code-generation",
"test-harness",
"multi-language",
"python",
"javascript",
"typescript",
"test-frameworks",
"parsing",
"file-io"
] | Add C++ and Rust targets (combined), exhaustive type support including `None/null`:
### Targets
`--lang cpp` and `--lang rust` are supported for `generate` and `test`.
### Value model parity with interpreted targets
All value types and behaviors supported in Python tests must be supported for C++/Rust targets too, ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for test_translator / checkpoint_6.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain babel_code_goat.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoi... |
test_translator | 7 | cpp | Hard | [
"code-generation",
"test-harness",
"multi-language",
"python",
"javascript",
"typescript",
"test-frameworks",
"parsing",
"file-io"
] | Async support across all target languages + timeouts and selection for `test`
### Async support across all targets
If tests use async/await semantics, the harness must run the entrypoint invocation to completion (or timeout) in every target language.
### `test` flags for selection + timeouts
`test` accepts:
* `--l... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for test_translator / checkpoint_7.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain babel_code_goat.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoi... |
test_translator | 8 | cpp | Hard | [
"code-generation",
"test-harness",
"multi-language",
"python",
"javascript",
"typescript",
"test-frameworks",
"parsing",
"file-io"
] | Introduce `profile`, with `-n`, `--warmup`, mean/std, memory profiling; flags parity with `test`
### `profile` command
`profile <tests_dir> <solution_path> --lang <target_lang> [flags...]`
`generate` must run first:
`profile` must error if the tester file for that language is missing.
### `profile` flags
`profile`... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for test_translator / checkpoint_8.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain babel_code_goat.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoi... |
textdrop | 1 | cpp | Easy | [
"http-server",
"markdown",
"storage",
"object-storage",
"rendering",
"cookies",
"health-checks",
"static-assets",
"lifecycle",
"startup-validation",
"failure-contracts"
] | ## TextDrop
TextDrop is a self-hosted text sharing service that accepts plain-text submissions via an HTML form, stores them under generated short URLs, and serves the stored content as rendered HTML pages or raw text.
The service runs as an HTTP server launched from the command line. Users compose text in a browser-... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for textdrop / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain textdrop.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
textdrop | 2 | cpp | Easy | [
"http-server",
"markdown",
"storage",
"object-storage",
"rendering",
"cookies",
"health-checks",
"static-assets",
"lifecycle",
"startup-validation",
"failure-contracts"
] | ## Markdown Rendering and Multi-Format Input
TextDrop is a self-hosted text sharing service that accepts plain text or markdown submissions, stores them, and serves rendered and raw views via short URLs.
The service provides markdown rendering and accepts JSON-encoded submissions in addition to form input.
### Markd... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for textdrop / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain textdrop.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
textdrop | 3 | cpp | Easy | [
"http-server",
"markdown",
"storage",
"object-storage",
"rendering",
"cookies",
"health-checks",
"static-assets",
"lifecycle",
"startup-validation",
"failure-contracts"
] | ## TextDrop — Metadata Extraction, OpenGraph Tags, Table of Contents, and Ephemeral Preview
TextDrop is a self-hosted text sharing service that accepts plain text or markdown submissions, stores them with generated short IDs, and serves rendered and raw views via short URLs.
The service extracts metadata from markdow... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for textdrop / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain textdrop.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
textdrop | 4 | cpp | Easy | [
"http-server",
"markdown",
"storage",
"object-storage",
"rendering",
"cookies",
"health-checks",
"static-assets",
"lifecycle",
"startup-validation",
"failure-contracts"
] | ## TextDrop — Static Content, Assets, and Crawl Policy
TextDrop is a self-hosted text sharing service that accepts plain text or markdown submissions, stores them with generated short IDs, and serves rendered and raw views via short URLs.
The service serves bundled documentation pages alongside user-generated notes, ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for textdrop / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain textdrop.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
textdrop | 5 | cpp | Easy | [
"http-server",
"markdown",
"storage",
"object-storage",
"rendering",
"cookies",
"health-checks",
"static-assets",
"lifecycle",
"startup-validation",
"failure-contracts"
] | ## TextDrop
TextDrop is a self-hosted text sharing service that accepts plain text or markdown submissions, stores them with generated short IDs, and serves rendered and raw views via short URLs. The service provides internal health aliases, authenticated lifecycle transitions, configurable operational secrets, and pe... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for textdrop / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain textdrop.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
textdrop | 6 | cpp | Easy | [
"http-server",
"markdown",
"storage",
"object-storage",
"rendering",
"cookies",
"health-checks",
"static-assets",
"lifecycle",
"startup-validation",
"failure-contracts"
] | ## Pluggable Storage Backends and Creation Metadata
TextDrop is a self-hosted text sharing service that accepts plain text or markdown submissions, stores them with generated short IDs, and serves rendered and raw views via short URLs.
The storage layer is configurable at startup, supporting both a local file-backed ... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for textdrop / checkpoint_6.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain textdrop.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning th... |
trajectory_api | 1 | cpp | Medium | [
"rest-api",
"json-validation",
"concurrency-control",
"data-aggregation",
"ebnf-parsing",
"sandboxed-execution",
"state-management",
"versioning",
"complex-validation"
] | Build a REST service to **store and manage agent trajectories from experiments**. Part 1 focuses on the **core API**: ingesting trajectories with strict validation, searching, and producing simple aggregate reports.
# Deliverable
An HTTP API server named **`trajectory_api.cpp`** that starts a REST service.
* It **mu... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for trajectory_api / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain trajectory_api.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
... |
trajectory_api | 2 | cpp | Medium | [
"rest-api",
"json-validation",
"concurrency-control",
"data-aggregation",
"ebnf-parsing",
"sandboxed-execution",
"state-management",
"versioning",
"complex-validation"
] | Extend your server to handle **in-progress trajectories** that can be **appended with new steps** and later **finalized**.
# Deliverable
Extend **`trajectory_api.cpp`** to support **mutable** trajectories.
* All new/changed endpoints below must be implemented.
* Mutations must use optimistic concurrency via the `If-... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for trajectory_api / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain trajectory_api.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
... |
trajectory_api | 3 | cpp | Medium | [
"rest-api",
"json-validation",
"concurrency-control",
"data-aggregation",
"ebnf-parsing",
"sandboxed-execution",
"state-management",
"versioning",
"complex-validation"
] | Now you must support **forking trajectories** to branch experiments from any prior step.
---
## Deliverable
Extend **`trajectory_api.cpp`** to support creating **forks** from an existing trajectory at a specified step, with optional immediate appends and metadata overrides.
* Implement the new/changed endpoints and... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for trajectory_api / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain trajectory_api.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
... |
trajectory_api | 4 | cpp | Medium | [
"rest-api",
"json-validation",
"concurrency-control",
"data-aggregation",
"ebnf-parsing",
"sandboxed-execution",
"state-management",
"versioning",
"complex-validation"
] | Here’s **Part 4**: add **tool-use tracking** by letting clients upload an **EBNF DSL grammar** for each environment. The server must parse assistant step content using the **active grammar version** of that environment and expose **exact tool-call extractions**. Only include what’s new or changed from Parts 1–3.
---
... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for trajectory_api / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain trajectory_api.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
... |
trajectory_api | 5 | cpp | Medium | [
"rest-api",
"json-validation",
"concurrency-control",
"data-aggregation",
"ebnf-parsing",
"sandboxed-execution",
"state-management",
"versioning",
"complex-validation"
] | Here’s **Part 5**: clients can upload **Python/Bash tool implementations** for an environment, and the server must **execute tool calls** (extracted in Part 4) **step-by-step** in a sandboxed working directory, tracking the **filesystem state after each assistant step**. Only include what’s new or changed from Parts 1–... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for trajectory_api / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain trajectory_api.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
... |
xjq | 1 | cpp | Easy | [
"cli-tools",
"xml",
"html",
"json",
"xpath",
"css-selectors",
"json-to-xml",
"data-transformation",
"query-language"
] | # XPath Querying of XML from Stdin
## CLI
Implement executable `xjq.cpp`:
`./xjq [OPTIONS] QUERY [INFILE]`
- `QUERY`: XPath expression.
- `INFILE`: accepted positional argument but not used.
- Input source: stdin.
## Behavior
- Parse stdin as XML/HTML using case-sensitive element matching.
- Evaluate `QUERY` with... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for xjq / checkpoint_1.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain xjq.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the problem ... |
xjq | 2 | cpp | Easy | [
"cli-tools",
"xml",
"html",
"json",
"xpath",
"css-selectors",
"json-to-xml",
"data-transformation",
"query-language"
] | # CSS Selectors and Text Extraction
## CLI Additions
- `--css`: interpret `QUERY` as CSS selector.
- `-t`, `--text`: direct text extraction from matched elements.
- `--text-all`: descendant text extraction from matched elements.
## CSS Query Mode
- Define custom `::text` pseudo-element:
`selector::text` returns d... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for xjq / checkpoint_2.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain xjq.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the problem ... |
xjq | 3 | cpp | Easy | [
"cli-tools",
"xml",
"html",
"json",
"xpath",
"css-selectors",
"json-to-xml",
"data-transformation",
"query-language"
] | # JSON Input and Auto-Detection
## Input Auto-Detection
- Auto-detect stdin as JSON object/array or XML/HTML.
- Top-level JSON primitives are not accepted as JSON input for this mode; they fall back to XML parsing.
- Input that fails JSON detection is parsed as XML/HTML.
## JSON-to-XML Contract
- Wrap converted doc... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for xjq / checkpoint_3.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain xjq.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the problem ... |
xjq | 4 | cpp | Easy | [
"cli-tools",
"xml",
"html",
"json",
"xpath",
"css-selectors",
"json-to-xml",
"data-transformation",
"query-language"
] | # File Input, First-Result, and Compact Output
## CLI Additions
- `-f`, `--first`: return only first result.
- `-c`, `--compact`: compact output mode.
## Input Source
- `INFILE` takes precedence over stdin when both are present.
- Additional positional args after `QUERY [INFILE]` are ignored.
- Decode `INFILE` as U... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for xjq / checkpoint_4.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain xjq.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the problem ... |
xjq | 5 | cpp | Easy | [
"cli-tools",
"xml",
"html",
"json",
"xpath",
"css-selectors",
"json-to-xml",
"data-transformation",
"query-language"
] | # Smart Output Formatting, JSON Export, and Multi-XPath
## CLI Addition
- `-j`, `--json`: JSON export mode for XML element results.
## Output Precedence
When multiple output flags are present:
1. `--text-all`
2. `--text`
3. `--json`
4. default auto-format behavior
Notes:
- `--json` has no effect when results are... | #!/usr/bin/env bash
set -euo pipefail
# SCBench test runner for xjq / checkpoint_5.
# Usage:
# bash test.sh /path/to/submission [extra pytest args...]
#
# The submission directory should contain xjq.cpp and, optionally,
# requirements.txt. Set SCB_PROBLEMS_REPO to an existing checkout to avoid
# cloning the problem ... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.